skill-creator

To create a Claude skill, you write a SKILL.md file with YAML frontmatter containing name and description fields that maximize discoverability. This frontmatter ensures correct triggering and reliable context loading for your specialized workflow.

Progressive disclosure in skill packaging is a token-efficient design approach where you keep core instructions concise in SKILL.md and move detailed information into bundled references, scripts, or assets. Use it to reduce context overhead while maintaining operational depth.

Building a Claude skill requires no external dependencies or frameworks. You structure the skill using a SKILL.md file with YAML frontmatter and organize supporting scripts, references, and assets directly within the skill bundle for token-efficient operation.

The best way to structure operational instructions in SKILL.md is to write concise, action-oriented guidance directly in the body while moving extended detail into bundled scripts and references. This progressive disclosure design maintains token efficiency and reliable context loading.

Your Claude skill may not trigger correctly if the YAML frontmatter name and description fields lack the vocabulary needed for discoverability. Clear, specific frontmatter aligned with your workflow automation domain ensures proper matching and context loading.